← Back to work

EdTech · Info Architecture · Research

Team Rubicon
Operations Platform

A 0 → 1 internal platform that unified disaster-response logistics for 1,000+ volunteer leaders, replacing five fragmented tools with one situational-awareness home.

ROLE

Sole Product Designer

TIMELINE

3 months ( Shipped )

TEAM

1 PM · 1 Eng Lead · Me

TYPE

0 → 1 · B2B

−42%

MEDIAN TASK TIME

Reduction in moderated tree testing across 12 representative tasks (n=18).

88%

FINDABILITY

Tree-test success rate after the second card sort, up from 51% baseline.

3 → 1

AUDIENCE INBOXES

Three audience-specific support channels consolidated into one shared queue.

OVERVIEW

The content wasn't broken — the structure was.

TSE's website served three very different audiences — students, faculty, and administrators — but treated them all the same. Navigation was disorganized. Key pages were buried 6 levels deep. Users bounced before finding what they needed. My challenge: restructure the entire experience without rewriting a single word of content.

HMW QUESTION

How might we help 3 distinct audiences find what they need — without confusion, friction, or drop-off?

3 audiences

Students, faculty, and administrators — each with entirely different goals and mental models

↑ 40%

Task completion rate increase across all three user groups in moderated usability testing

Zero new content

Every improvement came from restructuring nav, hierarchy, and labeling — not adding pages

RESEARCH

I had to see through three completely different lenses

I recruited 12 participants (4 per audience group) through the university's mailing lists. Using open card sorting via Optimal Workshop, participants grouped 47 page titles into categories that made sense to them. I then ran moderated tree tests to validate the proposed IA structure, measuring task completion rate and time-to-find for 6 critical tasks per audience.

Prospective Students

Admissions info, deadlines, program requirements

"I gave up and emailed instead." — Admission pages nested 4 levels deep. Couldn't distinguish between programs.

Faculty

Course tools, research resources, admin portals

"I've stopped trying to find things — I just search." — Bookmarked everything because navigation changed each semester.

Administrators

Policy docs, forms, deadlines, compliance tools

"Half our calls are people asking us to find things on our own website." — Forms weren't findable. Key deadlines lived in PDFs.

🖼

Attach Image

Card sorting affinity diagram — showing how 3 audiences grouped 47 pages differently

🖼

Attach Image

Tree test results — task completion rate before vs. after restructure

From 9 top-level items to 3 audience entry points

The original site organized 47 pages across 9 department-based navigation categories. After card sorting and tree testing with 12 participants, I restructured the entire IA around 3 audience-based entry points — reducing max depth from 6 levels to 3.

🖼

Attach Image

Before: 9-category sitemap (47 pages, 6 levels deep)

🖼

Attach Image

After: 3-audience sitemap (47 pages, 3 levels max)

PROCESS

Why I chose audience-first over department-first

The old site organized content by department — the way staff thought about it. I reorganized by task and audience — the way users actually think. Three focused entry points replaced one overloaded nav.

01 — IA

Audience-based nav structure

I replaced 9 top-level nav items with 3 role-based entry points. Users now route themselves immediately instead of scanning everything.

02 — Labels

Plain language over academic jargon

I replaced internal naming conventions with language users actually searched for. "Academic Affairs" became "Courses & Requirements." 22 pages relabeled.

03 — Hierarchy

Flatten depth, surface the critical

I reduced max nav depth from 6 levels to 3. Most-accessed pages promoted to top level. Deadline info surfaced on landing pages instead of buried in PDFs.

04 — Validation

Test with all three groups, not one

Each testing round included participants from all 3 audiences. A solution that helped students but confused administrators wasn't a solution — I had to keep all three lenses in focus.

IMPACT

Clarity is a design output, not a content problem

↑ 40%

Task completion rate improved from 45% to 85% across all three audience groups in moderated usability testing (n=12, 4 per group)

6 → 3

Nav depth for most critical tasks reduced from 6 clicks to 3 or fewer

22 pages

Relabeled using plain language: "Academic Affairs" → "Courses & Requirements," "Student Affairs" → "Campus Life & Support," "Office of the Registrar" → "Records & Enrollment"

REFLECTION

I initially assumed the problem was outdated content — that pages needed rewriting. But after testing with all three audiences, I discovered users could complete tasks with the exact same words, as long as those words were in the right place. My first IA draft grouped content by task type, but administrators kept failing because their mental model was role-based, not task-based. That insight pushed me to the audience-first structure. If I had more time, I would have conducted a quantitative tree test with a larger sample to validate the final IA statistically — the qualitative results were strong, but a broader dataset would have made the case even more compelling for stakeholders.

← Back to all work

Next: Borrow & Lend App

Buro · Mobile Design · Strategy · 2023

View next case study →

© 2025 Liwen Chang · Built with care & an opinion about IA